WASHINGTON – In what could prove to be the final word in the clinical controversy over the safety of prescribing digoxin in patients with atrial fibrillation, a secondary analysis of the roughly 18,000-patient ARISTOTLE trial has come down emphatically on the side of avoiding the venerable drug.
“The clinical implications of our analysis are that in the absence of randomized trial data showing its safety and efficacy, digoxin should generally not be prescribed for patients with atrial fibrillation, particularly if symptoms can be alleviated with other treatments. And in patients with atrial fibrillation already taking digoxin, monitoring its serum concentration may be important, targeting blood levels below 1.2 ng/mL,” Renato D. Lopes, MD, PhD, said at the annual meeting of the American College of Cardiology.
A randomized clinical trial of digoxin in AF is extremely unlikely, added Dr. Lopes, professor of medicine at Duke University in Durham, N.C.
ARISTOTLE was a randomized trial of apixaban (Eliquis) versus warfarin for stroke prevention in AF. The results of this landmark study, previously reported (N Engl J Med. 2011 Sep 15;365:981-92), demonstrated that apixaban was the superior oral anticoagulant in preventing stroke or systemic embolism, caused less bleeding, and resulted in lower mortality.
ARISTOTLE had some unique features that rendered the study database an exceptional resource for use in a large observational study of digoxin’s safety in patients with AF. It included a detailed serial assessment of concomitant medications as well as measurements of serum digoxin levels, left ventricular ejection fraction, creatinine clearance, and biomarkers including vasoactive intestinal peptide, troponins T and I, N-terminal pro–brain-type natriuretic peptide, and growth differentiation factor 15. These were among the 48 clinical variables included in multivariate adjusted analyses of mortality risk.
One-third of ARISTOTLE participants were on digoxin at study entry, a prevalence typical of what’s seen in clinical practice. Among the 5,824 subjects with AF already on digoxin at the start of the trial, the risk of death during follow-up proved independently related to baseline serum digoxin concentration. Patients with a level from 0.9 ng/mL to less than 1.2 ng/mL had a 16% increased risk of death during study follow-up, compared with digoxin nonusers, a trend that didn’t reach statistical significance. However, the 11% of AF patients with a serum concentration of 1.2 ng/mL or above were at a significant 56% increased risk for death.
When serum digoxin concentration is looked at as a continuous, rather than dichotomous variable, for each 0.5-ng/mL increase in drug concentration, the adjusted risk of all-cause mortality at 1 year of study follow-up climbed by 19%.
Moreover, among 781 AF patients who initiated digoxin during the study, the risk of death was increased by 78%, compared with that of 2,343 extensively matched controls. The most common cause of this excess mortality was sudden death, and in a closer look at that endpoint, the investigators found that the risk of sudden death was increased fourfold in new users of digoxin. This increased risk occurred early: Most sudden deaths occurred within the first 6 months after going on the drug, suggesting a causal relationship, although not providing definitive proof, Dr. Lopes noted.
Forty-three percent of ARISTOTLE participants had heart failure at enrollment. Interestingly, the increased risk of death associated with on-study initiation of digoxin was of similar magnitude, regardless of whether comorbid heart failure was present. The mortality risk was 58% greater in new users with heart failure, compared with matched nonusers with heart failure, and twofold greater in new users without heart failure than in their matched controls.
The benefits of apixaban over warfarin were consistent regardless of whether or not patients were on digoxin.
Discussant Kristen K. Patton, MD, was effusive in her response to the new ARISTOTLE findings.
“This was a really, truly, beautiful observational analysis,” declared Dr. Patton, an electrophysiologist at the University of Washington, Seattle.
“I think in cardiology, where our hearts have been broken before due to flawed observational studies, it’s really important for people to understand that observational data, when analyzed well, with appropriate propensity matching, with new-user analysis and close attention to clinical variables that are important, can really change practice in a good way. I think that’s what we see here,” she said.
A beaming Dr. Lopes responded that it’s likely that some of the past conflicting studies were marred by survival bias – that is, an inability to account for the fact that patients already on digoxin at the outset of a study have already declared themselves to be more tolerant of the drug. Past studies also didn’t adjust for biomarker levels.
“We could adjust for things we know today are associated with death in atrial fibrillation,” he observed.
Dr. Patton added that the most surprising study finding to her involved the new users of digoxin. She suspects that the reported figure of a 78% increased risk of all-cause mortality during study follow-up actually markedly underestimates the true size of that risk during the initial months on the drug. Dr. Lopes agreed.
She also said she found worrisome and disappointing the increased mortality risk reported with initiation of digoxin in AF patients with heart failure. That hasn’t been seen in other studies.
Dr. Lopes said the investigators utilized multiple means of identifying patients with heart failure and are certain they captured the full population of affected patients.
“We feel very confident that, when you have atrial fibrillation together with heart failure, it might be a different story than without atrial fibrillation,” the cardiologist said.
Discussant Jagmeet P. Singh, MD, associate chief of cardiology at Massachusetts General Hospital and professor of medicine at Harvard Medical School, Boston, said the ARISTOTLE analysis carries an eye-opening take-home message: “If you have to initiate digoxin, you have to follow the serum levels more closely than we ever have before. How frequently, I don’t know – maybe monthly instead of at the 6-monthly intervals that we often do. And I think maybe arrhythmia monitoring in the initial stages of putting patients on digoxin will be key to see if there are any additional proarrhythmic effects.”
The original ARISTOTLE trial was sponsored by Bristol-Myers Squibb and Pfizer. However, the ARISTOTLE digoxin analysis was sponsored by the Duke Clinical Research Institute. Dr. Lopes reported serving as a consultant to and/or receiving research grants from Bristol-Myers Squibb, Pfizer, Bayer, Boehringer Ingleheim, Daiichi Sankyo, GlaxoSmithKline, Medtronic, Merck, and Portola.